Improve error messaging for "Managed Images not supporting ARM64 images error"

[nywilken]

The issue lies on the use of managed_images for the compute gallery, which is not supported for this type of CPU. This behavior has been documented by our friends at Microsoft here.

By simply removing the managed imaged attributes from your build template I believe the build should work.

 "builders": [{
      "type": "azure-arm",

      "client_id": "{{user `client_id`}}",
      "client_secret": "{{user `client_secret`}}",
      "tenant_id": "{{user `tenant_id`}}",
      "subscription_id": "{{user `subscription_id`}}",
      "os_type": "Linux",
      "ssh_username": "rocky",
      "ssh_private_key_file": "/data/jenkins/.ssh/CLOUDTRUSTQA-01102018.pem",
      "disk_caching_type": "None",
      "os_disk_size_gb": "20",

    "shared_image_gallery": {
        "subscription": "{{user `subscription_id`}}",
        "resource_group": "{{user `resource_group`}}",
        "gallery_name": "{{user `gallery_name`}}",
        "image_name": "{{user `image_def`}}",
        "image_version": "{{user `IMAGEVER`}}"
   },

   "shared_image_gallery_destination": {
    "subscription": "{{user `subscription_id`}}",
    "resource_group": "{{user `resource_group`}}",
    "gallery_name": "{{user `gallery_name`}}",
    "image_name": "{{user `target_image_def`}}",
    "image_version": "{{user `Target_IMAGEVER`}}",
    "replication_regions": ["Canada Central", "Canada East"],

    "storage_account_type": "Standard_LRS"
},



      "azure_tags": {
      "NAME": "{{user `target_image_name`}}",
      "BUSINESSUNIT": "CICD",
      "BUSINESSENTITY": "CLOUDTRUST",
      "OWNEREMAIL" : "[email protected]",
      "RUNNINGSCHEDULE": "00:00:23:59:1-7",
      "APPLICATIONENV": "QA",
      "APPLICATIONROLE": "QA",
      "ALERTGROUP": "ops_team",
      "SERVICENAME": "packer",
      "RELEASE": "{{user `release`}}",
      "POD": "NOPOD"
      },

      "location": "westus2",
      "vm_size": "Standard_D2pds_v5"
    }
    ],

@JenGoldstrich did a better job at explaining the issue here, which can may be helpful to you in troubleshooting the error.

Originally posted by @nywilken in #366 (comment)

[JenGoldstrich]

Just a thought maybe it would be better to see if its feasible to check the image architecture before deploy, so that we can error early without getting to the deployment of the managed image, I imagine this is doable but I haven't investigated it further